Time-lapse acoustic impedance (AI) inversion for Weyburn oil field (southeastern Saskatchewan, Canada) is performed by combining the data from seismic records, well logs and velocities inferred during reflection seismic processing. The objective of time-lapse AI mapping in the field consists in constraining fluid migration during CO2 injection and enhanced oil recovery. The time-lapse AI is derived by a three-step procedure: 1) time-variant seismic calibration of the records from three available vintages of the dataset; 2) evaluation of the differential reflectivity and 3) derivation of the differential AI from differential reflectivity. Although applied to early stages of CO2 injection, the resulting differential AI images and time-shift images suggest indications of CO2 migration.
Presentation Date: Thursday, September 28, 2017
Start Time: 10:35 AM
Presentation Type: ORAL
We present a synthetic data study to investigate a number of causes of noise in time-lapse seismic experiments, including water velocity variations, source and receiver repeatability, and coordinate accuracy. Synthetic oceanbottom node datasets were generated for base and monitor surveys with realistic acquisition parameters, and 4D difference images produced to study each source of noise independently. Results indicate that factors that can be most controlled during acquisition (repeatability) contribute the least to overall 4D noise. Realistically achievable node repeatability is fairly good and produces the smallest amount of noise, while sources, which are the least repeatable, produce almost three times the amount of noise in our experiment. However, variable water velocities (not corrected for during processing) produce the majority of residual noise in 4D difference images.
Ghaderi and Landrø (2011) gave an example of how capillary pressure effects might cause zonation in the CO2 saturation for an example studying storage of CO2 in a subsurface sand layer. In this paper we will estimate average saturation versus hydrocarbon column thickness for oil and gas. Since the capillary pressure is directly related to the density difference between water and the hydrocarbon pore fluid, the saturation distribution will also be different, and especially when the hydrocarbon column get thinner towards the rim, these differences might be significant. In some exploration settings, we think it might be possible to estimate amplitude slopes directly from the seismic data, and use differences in these slopes to discriminate between oil and gas. In hydrocarbon exploration it is challenging to discriminate between oil and gas, and to study the slope of the amplitude behavior versus hydrocarbon thickness close to the rim, might become an additional exploration tool.
Søndenå (1991) found the following empirical relation between capillary pressure (PC , measured in bar) and water saturation (SW):
So far, oil and gas production in the deepwater Gulf of Mexico is mostly from Neogene (Pleistocene, Pliocene, and Upper Miocene) reservoirs. The Neogene reservoirs can be characterized broadly as overpressured, unconsolidated, and highly compacting, with high permeability and containing black undersaturated oil of medium gravity with moderate gas/oil ratio and some aquifer support. Although waterflooding is a mature technology, few water-injection projects have been conducted in the Neogene reservoirs because they exhibit good primary recoveries, exist in high-cost offshore environments, and are relatively small. In some of these fields, a limited volume of water was injected.
Waterflooding can supply additional reservoir energy for producing substantial quantities of oil trapped due to limited displacement drive and poor sweep efficiency. However, water injection is not commonly used in the deepwater Gulf of Mexico (DW GoM) due to good primary recovery, drilling cost and facility limitations. In over 80 fields and 450 reservoirs, water injection program has been implemented in only 18 reservoirs in 13 fields, or less than 5% of potential waterflooding candidates.
DW GoM mid-Miocene reservoirs are characterized by sparse well counts, over-pressured, and generally good rock and fluid properties. Rock compaction and moderate aquifer influx often provide moderate to good natural drive energy and oil recovery. Primary oil recovery averages 32% with the 80% confidence range between 16% and 48%. However, Paleogene reservoirs are characterized by deeper depth, high pressure, high temperature, complex geology, and rock and fluid properties. Estimated recoverable oil is only 10% of OOIP assuming primary production and limited natural drive energy. Water injection programs will be difficult to execute in tight, abnormally-pressured Paleogene reservoirs. Waterflooding of deepwater turbidites has accumulated many lessons and learns now, and a comprehensive understanding of the influence of depositional environment and injection into over-pressured, highly compacting rocks is necessary. This paper is a detailed examination of Pleistocene-to-Upper Miocene age turbidite reservoirs in the DW GoM under water injection. Issues on waterflooding these deepwater plays were reviewed in the context of geological setting and depositional environment. Despite many drawbacks that tend to oppose the implementation of a waterflooding in Paleogene reservoirs, this paper still proves that they are candidates for water injection programs under the rules of good production practice. Moderate oil recovery is suggested in highly compacting reservoirs with supplemental injection drive. Overall, waterflooding strategies have proven to be highly effective in achieving good incremental oil recovery from the deepwater Gulf of Mexico reservoirs.
Oil and gas producers have shown renewed interest in developing reservoirslocated both onshore and offshore within the Arctic regions of Alaska, Canadaand Russia. In many cases, the hydrocarbon reservoirs are known to be overlainby a massive permafrost interval that extends over depths of up to 700 m belowthe surface active layer. These conditions create unique design and operationalchallenges for production and injection wells from the perspective of ensuringthat well integrity will not be compromised by the inevitable thaw subsidenceof the permafrost soil layers.
The permafrost soil layers surrounding arctic wells will thaw gradually withtime due to wellbore heat loss. As the thaw zone advances radially outward fromeach well, the ice-to-water phase change within the pore space of thefrozen/partially frozen sediments will lead to changes in the permafrost soilproperties and to the loading conditions within the thaw column region. Thesechanges will result in soil deformations (including both vertical settlements(subsidence) and horizontal displacements) which can, in turn, inducesignificant well casing strains that need to be considered in selecting thewell design and layout. The magnitude of the soil deformations that occurthroughout the permafrost interval are highly dependent on the depositionhistory, insitu temperature and the physical and mechanical properties of theindividual soil layers. Therefore, in order to accurately predict the soildeformations and resultant localized casing strain levels, it is essential toobtain reliable data to properly characterize the lithology (soil types) withinthe permafrost interval, as well as the frozen state and the relevantmechanical and thermal properties (both frozen and thawed) of individual soillayers. This paper describes the various information and geotechnical test datathat has been used to establish the thaw and deformation response of differentpermafrost soils at a number of arctic locations for the purpose of evaluatingthe effects of thaw subsidence loading on wells. Overall, the paper serves tohighlight the importance of collecting the appropriate geotechnical data toallow thaw subsidence-induced ground deformations and associated casing loadingconditions to be properly considered at the well/project design stage.
The two issues of source signature repeatability and source array directivity have been considered in great detail by many authors. Loveridge
Behrens, Ronald (ChevronTexaco E&P Technology Co.) | Condon, Patrick (ChevronTexaco E&P Technology Co.) | Haworth, William (ChevronTexaco North America Upstream) | Bergeron, Mark (ChevronTexaco North America Upstream) | Ecker, Christine (ChevronTexaco E&P Technology Co.)
Many waterflooded or aquifer influx reservoirs are old or marginal and thus do not have ideal data conditions, either historic or current, for 4D seismic. We modeled the synthetic seismic response for the 7100 Sand in the Bay Marchand field and found it suitable for 4D. To determine the noise level in modeling, we derived relations for calculating the appropriate level of noise from differences in the field data. We also developed relations between measured noise and the quantitativeness of calculated saturation changes. These methods, as applied to the Bay Marchand, showed that useful saturation predictions could be made from the seismic data shot 30 years after initial production with a recent, nonexclusive survey. From these data, both downdip and updip wells were drilled successfully.
Multiple seismic surveys over time that were not intentionally designed for 4D evaluation can be used for imaging fluid and pressure changes and locating bypassed reserves. Bay Marchand is a mature Gulf of Mexico field with a series of stacked reservoirs separated by shale. The first 3D seismic survey was acquired after first production, and the second seismic survey was a nonexclusive survey having very different acquisition parameters. These circumstances make interpreting 4D results difficult but not impossible. We were able to make a reasonable interpretation of the "A" fault block on the original data before cross-equalization (XEQ), although XEQ reduced the acquisition and processing differences between the cubes. We were able to identify areas of water influx and areas not yet invaded by water, even though we were prevented from following the typical steps in 4D. Reliability estimates were derived for estimates of saturation change as a function of signal-to-noise ratio and scatter in the rock physics relationships. This paper discusses these steps in contrasting the ideal case with Bay Marchand and shows where a 4D project can still be successful even with compromises forced by lack of early data or current budget.
Using Time-Lapse Seismic Monitoring
The potential benefits of time-lapse (or 4D) seismic monitoring have been well recognized across the industry worldwide over the past 5 years.1 It has been applied to a number of fields in the Gulf of Mexico.2,3 The emphasis toward permanent monitoring of the reservoir and the concept of the instrumented oilfield show the true potential of 4D as a tool for reservoir management. But there are many fields that began producing before the recognition of the benefits of 4D. In those fields, the data are often not optimally acquired and may not even be available preproduction. In mature fields, the economic environment is driven by cost, which in turn impacts the data that can justifiably be acquired.
The key elements of a successful 4D seismic project consist of feasibility,4 acquisition, processing and interpretation.5 Feasibility comprises two factors: detectability and repeatability. Detectability is the amount of change in the elastic properties of the reservoir associated with production. That change might be from fluid saturation, pressure, temperature, or a combination. In specific cases, production-related changes also may cause changes in porosity. Construction of an appropriate rock physics model is a critical element to assessing detectability. The key data for this purpose are core measurements and/or well logs. Repeatability is a measure of the similarity of the seismic response between two or more seismic surveys (excluding production-related effects). Unwanted differences in the amplitude and timing of seismic reflections can be created by differences in seismic acquisition and processing. To interpret amplitude and time differences created by the changes in the reservoir properties, it is necessary to minimize these effects.6,7 There is a tradeoff between these two factors so that low values of detectability require a high value of repeatability. On the other hand, seismic data with a low value of repeatability may still be of value if the detectability is sufficiently high. This low repeatability is typical of legacy 4D seismic projects in which two 3D seismic surveys are available, but for which acquisition and processing of the data are not identical. This is the case described here. Bay Marchand fits the low repeatability/high detectability case.
Interpretation of 4D seismic data also can be subdivided into the qualitative and the quantitative.8 Qualitative interpretation recognizes where seismically detectable change is happening, but it only infers the significance of those changes. Qualitative changes may be the basis for infill drilling if the inferences are consistent with the rock physics model. Quantitative interpretation attempts to resolve those changes in seismic properties into changes in reservoir properties. Here, the rock physics model becomes the basis for converting seismic properties to reservoir properties. Ultimately, a quantitative interpretation might prove the basis for updating the simulation model.
This paper uses Bay Marchand as a case study to illustrate the impact of 4D on mature field development with seismic data obtained only during production and with unavoidable differences in seismic acquisition and processing. A description of the reservoirs, the geologic modeling, and the seismic modeling and analysis is given first to set the stage for the field results. The analysis section shows that the rock physics are very favorable to 4D, which compensates for the differences in seismic acquisition and processing. The interpretation stage includes understanding the seismic repeatability and how much it limits the quantitative or qualitative interpretation. We define a method for incorporating the seismic repeatability and the scatter in the rock physics to determine the nature of the interpretation. Combining well and wireline results confirms the reliability of the qualitative interpretation at Bay Marchand. The last section describes the time-lapse seismic interpretation, how it fits with the wells, and the results of the wells drilled on the basis of the 4D.